reduce the tokens

app.py

@@ -16,7 +16,7 @@ model_id = "unsloth/Llama-3.2-11B-Vision"
 model = AutoModelForImageTextToText.from_pretrained(
     model_id,
     torch_dtype=torch.bfloat16,
-    # device_map="auto",
+    device_map="auto",
 )
 processor = AutoProcessor.from_pretrained(model_id)
 
@@ -32,7 +32,7 @@ def score_it(input_img):
     prompt = "<|image|><|begin_of_text|>extract the text in this picture"
     inputs = processor(image, prompt, return_tensors="pt").to(model.device)
 
-    output = model.generate(**inputs, max_new_tokens=300)
+    output = model.generate(**inputs, max_new_tokens=200)
     return processor.decode(output[0])